Research Journal of Applied Sciences, Engineering and Technology 5(22): 5170-5181, 2013
ISSN: 2040-7459; e-ISSN: 2040-7467
© Maxwell Scientific Organization, 2013
Submitted: July 24, 2012
Accepted: August 21, 2012
Published: May 25, 2013
Response Surface Methodology in-Cooperating Embedded System for
Bus’s Route Optimization
1
Muhammad Rauf, 1Ahmed N. Abdalla, 1Azhar Fakharuddin and 2Elisha Tadiwa
1
Faculty of Electrical and Electronics, University Malaysia Pahang, Pekan, Pahang, 26300,
2
Computer Information System, University Technology Petronas, Tronoh, Perak, 31750, Malaysia
Abstract: The incessant increment of the population and lack of the reliable bus position notification system in big
cities is causing the boost of the private owned cars, has reasoned traffic congestion problems. A reliable public
transport notification system for passengers is able to resolve the issue stated above. First ever time, Response
Surface Methodology (RSM) is adopted in this study to presents a new real time intelligent buses notification system
using routes’ coordinates optimization and wireless mesh network emerged in an embedded design. The von Mises
and displacement equation is obtained through RSM and trained to the dedicated controller unit. A mature
experimental setup is designed to analyze the equation validation. The result from statistical and experimental
analysis using RSM implies that the target to optimize a single route coordinates into one digit representing bus stop
number is achieved by ~98%. The outcomes of research as location data minimizing and transmission time
reduction have a promised potential to a better and reliable public transport notification system using route’s
coordinates optimization.
Keywords: Embedded system, intelligent transportation system, optimization, radio transceivers
INTRODUCTION
The lack of fideism upon the public transport
infrastructure led the travelers to shift to their own
private car. The requirement of a reliable public
transport monitoring structure is the outcome from the
unnecessary congestions of traffic observed due to
steep increment in personal vehicle. A large scale
dedicated monitoring infrastructure with reliable
displaying position for public transport can facilitate
passengers to revolutionize the trend from own vehicle
to public transport as well international image of a city
(United States Government Accountability Office
(GAO), 2009; GR Reporter, 2011).
Intelligent Transportation System (ITS) has
witnessed tremendous growth over last few years. It is
adopting implementation of newest technology,
specifically a network of sensors, microchips, and
communication devices that collect and disseminate
information about the functioning of the transportation
system (Ban et al., 2011; Ezell, 2010). In the past, the
traffic monitoring infrastructure had mainly consisted
of dedicated equipment, such as loop detectors,
cameras, and radars. Later on Installation and
maintenance expenses discouraged the deployment of
these technologies for the entire transport network.
Moreover, inductive loop detectors are prone to errors
and malfunctioning (daily in California, 30% out of
25,000 detectors do not work properly according to the
PeMS system) (Herrera et al., 2010).
Through the advent of GPS technology,
conventional positioning methods have been replaced.
As far as the vehicle fleet management and monitoring
is concerned, one of the main applications of the GPS
technology is the fleet monitoring in urban or suburban
areas, from a central monitoring station (reference
station) (Mintsis et al., 2004; Lee and Han, 2003).
Moreover, the location data transmission link had also
been the subject of keen interest for several researches.
The numbers of studies are found using GPRS for the
vehicle monitoring to an IP fix control center. In
cooperating GPRS and other data transmission
techniques like GSM, RFID and Bluetooth, the vast
utilization of embedded system is observed as well for
receiving, matching and transmitting location data (Bao
et al., 2002; Niu et al., 2009; Qin et al., 2008; Lu et al.,
2007; Soo et al., 2007; Bong et al., 2007). From the
contrast of past tool for efficient communication, GSM
based monitoring system seems to be the better research
tool (Herrera et al., 2010; Soo et al., 2007). However,
the cellular network based monitoring system for public
transport may cause various drawbacks including
unavailability of service, server down, no network
coverage
and
cost
consuming
even
after
implementation. In a review paper by Xuesong et al.
(2008), it is intend to evaluate the technical feasibility
of applying emerging wireless network technologies for
Corresponding Author: Muhammad Rauf, Faculty of Electrical and Electronics, University Malaysia Pahang, Pekan, Pahang,
26300, Malaysia
5170
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
resources tracking at building construction sites. Based
on critical reviews of available localization methods
and
state-of-the-art
wireless
communication
technologies, it is identified that the Zig-Bee-based
wireless sensor network and the Received Signal
Strength Indicator (RSSI) method are most promising
for solving on-site resources tracking problems.
Numerous researches can be observed on the
utilization of intelligence techniques to enhance the
visibility and observability of GPS. The navigational
errors reduction like multipath amplitude delay and
phase relative to the direct path are well documented in
Liu and Amin (2009). Another approach of neural
network integration with the kalman filter based
optimization techniques on the GPS design and its
application for better utilization in the area of vehicular
navigation was presented (Fan and Jiancheng, 2007).
However, INS/GPS integration techniques using
kalman filtiring, have some inadequacies related to the
stochastic error models of inertial sensors, immunity to
noise, and observability. In the perspective, Sharaf and
Noureldin (2007) aimed to introduce multi sensor
system integration approach for fusing data from INS
and GPS for the navigation purpose, utilizing Artificial
Neural Networks (ANN). A multi layer perceptron
ANN has been recently suggested to fuse data from INS
and Differential GPS (DGPS). Recently, methods based
on Artificial Intelligence (AI) like Input-Delayed
Neural Networks (IDNN) have been suggested for
integrating the Global Positioning System (GPS) with
the Inertial Navigation System (INS) (Noureldin et al.,
2011).
Regarding other optimization techniques, Design
of Experiments (DOE) had been a useful tool that used
for exploring new processes, gaining increased
knowledge of the exiting processes and optimizing
these processes to achieving a better performance
(Rowland and Antony, 2003). Dynamic and foremost
important tool of Design of Experiment (DOE),
wherein the relationship between responses of a process
with its input decision variables is mapped to achieve
the objective of maximization or minimization of the
response properties (Raymond and Douglas, 2002).
The related art and literature evidently describes
that in the vast domains of navigation system for
transport, the crucial issues are the consistent
monitoring system with reliable error compensation
using maximum faultless techniques. In order to
challenge past exertions, this study focus on developing
a new intelligent multi ID buses monitoring system for
passengers. For the purpose, first field experiment was
conducted in the Kuala Lumpur city of Malaysia to
record the static coordinates for bus stops of single
route to create suitable input factors for RSM, the
response
methodology
of
DOE.
Regarding
technological approaches, the improved accuracy of
GPS technology and long range Zigbee standards has
seemingly become as the preferred choice of location
detection and transmission method for ‘high-end’
system because of their compatibility with prevalent
used embedded system design. Technically, the
comprehensive proposed system comprises of two
major modules; one is master module (the buses) while
second one is named slave module (the bus stations).
For the efficient location data forwarding, looping of
the radio chips, impressed by the concept of wireless
sensor networking is applied. This location data
transmission is made for the best visualization result at
each station node. PROTON+ IDE and DOE version 8
are chosen as software research tools. First time, in the
Fig. 1: Conceptual frame work of system
5171
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
field of ITS, DOE is selected to utilized. Response
Surface Method (RSM) is used to estimate the response
at the optimal region. The estimated function is then
used to optimize the responses. Similar to the factorial
design, linear regression and ANOVA are the tools for
data analysis in RSM. Hence Central Composite Design
(CCD) approach from RSM was selected for the present
study in order to shape out the response of the fixed
coordinates into a single linear equation. Later on the
equation is trained to the controller unit for efficient
and improved communication link between vehicle and
its related infrastructure.
stop X and Y are found in front each other and there is
a sound possibility for Y to receive the signal from bus
“A”. This problem is figure out by giving dedicated
PAN ID same as the bus ID. Let suppose, the bus A
have its ID ‘B103’, then transceiver modules of bus
“A” and its stops W and X are configured “0103” as
PAN ID. Through this scheme, the bus is made
responsible to communicate or to send its location
information to its relevant bus stops, not even the one
found in front or in range.
CONCEPTUAL FRAMEWORK
Given the lack of an established body of research
on the design of public transport network services and
the small number of practitioners concerned in this
field, a comprehensive approach seems to be required.
For the purpose, initially some rules and regulations are
described as the initial system assumptions. These
assumptions are described to make the proposed frame
work as resourceful as possible. Second is the hardware
implementable design, the center of focus of this study,
for the overall system to work. This phase contains best
functional components selection according to the
proposed design.
Real-time traffic notification system for virtually
every bus stop in the city has the ability to change the
trend of a traveler from using their personal vehicle to
the single public transport. During an informal survey
to the roads of KL, many problems are commonly
observed which must consider during framework
designing. Like many time buses having same IDs draw
closer together. Some time bus stops are found in front
of each other across the road. The conceptual solution
suggested for the difficulties found, are the scheming of
our system using GPS based monitoring and their
location transmission down to the specific bus stops
only found on their routes, not even at the bus stop
found across the road as exposed in Fig. 1. This specific
location transmission is actually the function of
specially configured transceivers and initially inputs by
the drivers. The earlier version of this conceptual
framework can be found at Rauf et al. (2010) and Abd
Alla et al. (2011).
In order to elaborate the conceptual framework, an
example is taken into account containing three buses
with different routes named A, B and C and 6 bus stops
with different route location named U, V, W, X, Y and
Z as shown in Fig. 1. Bus ID “A” will pass through bus
stop W and X. It can be observed from figure that bus
Table 1: Bus stops coordinate data (Bus-B103)
Received coordinates
Name of bus stations (slave node)
Longitude
--------------------------------Deg.
Min.
Sec.
Jalan Ipoh
101
41
42.44
• Hospital Damai (Pi1)
101
41
55.51
• Stesen Monorel Chow Kit (Pi2)
Jalan Raja Muda Abd Aziz
101
42
11.21
• Stadium Raja Muda (Pi3)
101
42
43.02
• Menara TH Selborn (Pi4)
Jalan Tun Razak
101
42
58.45
• City Square (Pi5)
Jalan Ampang
101
43
03.50
• Hotel Nikko (Pi6)
101
42
45.48
• KLCC (Pi7)
101
42
19.08
• Hotel Renaissance (Pi8)
Jalan Sultan Ismail
101
42
20.94
• Hotel Concorde (Pi9)
Jalan P Ramlee
101
42
26.60
• Menara KL(Pi10)
101
42
30.21
• Wisma Lim Foo Yong (Pi11)
FRAMEWORK DESIGNING
Assumptions: The frame work presented in the
methodology below, is based on the following some
assumptions must be imposed by the regulators:
•
Each bus route is isolated entities, are assigned
similar activities
The inter-linked network of nodes represents the
service along a bus route and hence the efficiency
of the network is equivalent to that of the service
provided along the route
Buses of same IDs are sent by a proper interval of
time. In case, any two or more buses of same IDs
join each other due to any reason then 2nd one
need to stop for the time until first one reach at the
next bus stations
•
•
Trained coordinates
Latitude
----------------------------Deg.
Min.
Sec.
Longitude
-----------------------Min.
Sec.
Latitude
-----------------Min.
Sec.
Bus stop no.
03
03
10
10
12.36
03.11
41
41
42.4
55.5
10
10
12.3
03.1
1
2
03
03
10
10
04.83
04.71
42
42
11.2
43.0
10
10
04.8
04.7
3
4
03
09
50.41
42
58.4
09
50.4
5
03
03
03
09
09
09
36.57
31.79
23.84
43
42
42
03.5
45.4
19.0
09
09
09
36.5
31.7
23.8
6
7
8
03
09
18.55
42
20.9
09
18.5
9
03
03
09
09
13.10
01.93
42
42
26.6
30.2
09
09
13.1
01.9
10
11
5172
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
•
Buses' driver are strictly ensured not to change the
predefined route. In case any emergency, they are
restricted to inform to the monitoring room so that
the monitoring plan could be change to facilitate
the passengers
System description: The design assures the real time
processing of data related to the position monitoring of
bus and more towards its implementation through
notification system for passengers. Complexity of
overall system is ensured to reduce by dividing it into
two basic parts; master part and slave part 1, 2 and so
on, as described in the concept, depending on number
of bus stations.
Master module:
Master module integration: Given the design needs
and requirement for the problem, it is proposed to come
up with the simplest and efficient design that would be
able to forecast the real time localization down to the
respective node. In order to construct an appropriate
design of master module, several best functioning
components are selected for lab and field test. In this
section of methodology, there are three key modules
used in master node: MCU: PIC16f877A, GPS module:
Parallex RXM-SG GPS Module and the advance
wireless: Xbee pro RF Modules (IEEE802.15.4). In the
integrated system, RSM is adopted to minimize the
location data for better, reliable data and time saving
comparison and transmission. On the basis of
prescribed dedicated equipment, standard prototype
system architecture was build to present optimized
experimental results. Two different experiments were
conducted to exploit the modules. Firstly the equation
obtained from RSM is validated using the master
module. The efficient outcomes from the proposed
heuristic will be shown in results section. Second is to
analyze the one way communication for sending bus
unique ID and one digit bus stop number from the
master node to the nearest determined slave node and
further more as looping. This determination is carried
out by assigning a unique PAN ID (a number shared
amongst each Xbee in a network) to the XBEEs of
master and three slave module prototype (representing
single route with its relevant bus). A single network
may contain many more nodes since XBEEs on
different networks do not communicate with each other.
The configurations of the RF modules are carried out
using X-CTU software. It allows setting all of the
parameters using a Graphical User Interface.
varying distance testing. It may play a considerable role
against the resource full communication and real time
implementation of the system. There are found many
reasons beside like dense urban traffic, long buildings
and atmospheric effect on data rate and reliability.
Besides, there are also found some error possibilities of
exact MM comparison, since the master node (bus) is
observed, do not stop on the predefined slave node (bus
stops). This lead the research to create idea of reduced
amount of the location data to be compared and
forward. To tackle this issue, this work is inspired by
the fact that vehicles move along routes with a known
map of coordinates.
In order to train the controller, a survey is carried
out to observe and record the location data by using the
acquisition utility of handheld GPS while driving the
vehicle around the city of Kuala Lumpur, Malaysia. In
all, approximately 40 min of driving, data were
collected over 11 set of coordinates from bus stops of
specific bus. The map and Table 1 shows the bus stops
chosen to collect static position data. From the casual
observation, it is found that during one full journey
from first until last stop, there are few digits from the
longitude and latitude are subject to be change while
some remained unchanged. The Table 1 shows the
actual static location address of bus stops. For an
autonomous passive localization transmission loop, all
together 05 significant digits from longitude and
latitude will be decided to select as the input for the
RSM. The apprehension behind this selection was the
consistency of the degree digits throughout one full
journey and last bit of second cast no any affect on
representing the bus stop number. The coordinate’s data
is collect from bus route B103.
Data pre-processing: Prior to use data from Table 1
for experimental testing, pre-processing was needed to
generate data sets containing the necessary information
to calculate the bus stop number. In light of the
screening observation, it is selected to study the effects
of the four significant factors, namely; longitude M,
longitude S, latitude M and latitude S. Furthermore,
parking violation on the pre identified bus stops to pick
and drop passengers, have potential to produce
unwanted outputs or malfunctioning for MM. The
parking at bus stops is crucial part of the study for
having proper MM and triggering of RF modules
dedicated to bus stops in the loop. In order to avoid the
error due to mismatching, each selected data set of
coordinate are fragmented in to three more possibilities.
The possibilities are recorded by observing the location
coordinates just 15 meters before and after the exact
Data collection for master module: Subsequent to the
bus stop. These possibilities are solely selected to
successful location data reception through GPS mount
utilize as input to get the response as predicted one digit
on master module, it is observed lack of exact data
bus stop number. The four variables, their levels and
forwarding, missing or time delay using preliminary
possibilities are listed in Table 2.
5173
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Table 2: Processed possibilities of coordinate data for RSM
Received coordinates
Longitude
Latitude
---------------------------------------Case
Min.
Sec.
Min.
Sec.
(P i 1,1 )
41
42.2
10
12.4
(P i 1,2 )
41
42.4
10
12.3
(P i 1,3 )
41
42.5
10
12.2
(P i 2,1 )
41
55.2
10
03.4
(P i 2,2 )
41
55.5
10
03.1
(P i 2,3 )
41
55.7
10
03.0
(P i 3,1 )
42
11.0
10
04.9
(P i 3,2 )
42
11.2
10
04.8
(P i 3,3 )
42
11.5
10
04.8
(P i 4,1 )
42
42.6
10
04.7
(P i 4,2 )
42
43.0
10
04.7
(P i 4,3 )
42
43.3
10
04.7
(P i 5,1 )
42
58.2
09
50.7
(P i 5,2 )
42
58.4
09
50.4
(P i 5,3 )
42
58.7
09
50.1
(P i 6,1 )
43
03.3
09
36.8
(P i 6,2 )
43
03.5
09
36.5
(P i 6,3 )
43
03.8
09
36.3
(P i 7,1 )
42
45.1
09
31.5
(P i 7,2 )
42
45.4
09
31.7
(P i 7,3 )
42
45.8
09
31.9
(P i 8,1 )
42
18.6
09
24.1
(P i 8,2 )
42
19.0
09
23.8
(P i 8,3 )
42
19.2
09
23.5
(P i 9,1 )
42
20.6
09
18.9
(P i 9,2 )
42
20.9
09
18.5
(P i 9,3 )
42
21.1
09
18.3
(P i10,1 )
42
26.3
09
12.8
(P i10,2 )
42
26.6
09
13.1
(P i10,3 )
42
26.8
09
13.5
(P i 11,1 )
42
30.0
09
02.3
(P i 11,2 )
42
30.2
09
01.9
(P i 11,3 )
42
30.7
09
01.6
(P i l, m ):
P = Platform or bus station; i = Identical network ID for
each bus route (PAN address); l = No. of bus stops (Slave modules);
m = No. of possibilities of particular bus stoppage (master module)
Table 3: Configuring parameters for conducting loop test
For bus route B103 Xbee # 1
Xbee # 2
Baud rate
(3) 9600
(3) 9600
Pan Id
0103
0103
Atmy
10
11
Atdl
11
12
Xbee # 3
(3) 9600
0103
12
13
Surface response methodology: Considering RSM in
the field of ITS for the first ever time, there are several
steps needed to cogitate using the data from Table 2, to
come up with required results. As mentioned earlier, the
four (04) major factors named as longitudeM,
LongitudeS, LatitudeM and LatitudeS are chosen in
order to achieve their effects on the response as
predefined output. The value range of the factor is
determined by the range of coordinates from first bus
stop until the 11th bus stop, selected as array. This
design matrix CCD is selected to describe the
experiment in terms of the actual values of factors and
the test sequence of factor combinations. In this
research, statistical method ANOVA is applied as the
tools for data analysis. The clear objective for all the
steps is to achieve a simplified mathematical response
of the system. Recommendations and ANOVA are
provided in results section.
Master module system flow: The GPS data acquisition
utility was written in the PROTON IDE to interface the
GPS receivers with the MCU as well as to compute the
bus stop number for comparison. The GPS receivers
output ASCII sentences every second using the NMEA
0183 serial communication standard. Subsequent to
receive the position frame, MCU is programmed in
such a way that only few significant digits (Table 2)
will be allow for the map matching calculations to
come up with the single digit bus stop number. The
exact conditional approach is applied for comparison
the bus stop number corresponds to its location address.
On successful comparison, the real time response using
equation will trigger the RF module. In short, all the
processing tasks for master module can be generally
distributed into three tasks, first is GPS location data
reception, second is bus stop number computing using
equation then third is transmitting the bus stop number
to the first slave node.
Slave module: Subsequent to the successful reception
of bus stop number to the respective node, special
algorithms are designed to operate the number of slave
modules corresponding to the number of bus stations.
In order to develop a reliable communication link
between master and its all respective slave module, it is
decided to include looping concept of sensor nodes
inspired by WSN. Each sensor node is also capable of
receiving as well as transmitting the position output
trigger data to the next node found on the same network
only. The least number is slave nodes are kept equal to
the maximum number of bus stops found on the way of
particular bus ID. For testing and validation purpose, a
set of three slave nodes are prepared to test the proper
looping and distance between three nodes. The Table 3
shows the configuration parameters for efficient
looping test between three slave node prototypes.
The PAN and Destination Addresses (DA) of the
selected transceiver modules are configured by X-CTU
software, so that a proficient looping of location until
the last bus station could made possible The
methodology used for selecting the destination and
source is kept such a way that the destination address of
first slave module is configured same as the source
address of successive second salve module. The same
destination and source address configuration technique
is adopted between second and third. Successful testing
of three modules showed the potential methodology
application to any number of slave modules.
Initially while the bus (master module) is ready to
move, it is programmed to configure as just transmitter
first and the nearest bus station (slave module 1) will be
configured as receiver first to send the bus unique ID.
Subsequently, the master module will start receiving
GPS signals for the bus location then after calculating
5174
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Fig. 2: Overall system integration
Fig. 3: Overall process flow
and matching, will literally start to transmit to the first
nearest bus station. When the bus is on the move, the
slave module 1 will be configured as transceiver and
will send the one digit bus stop position to slave Module 2
(2nd next bus station) and up to so on. This chain for
receiving, displaying and transmitting bus positioning
information will be last till the final bus station for
every bus found in the particular route. The same flow
will be followed while the bus would pass through its
relevant bus stops. For the prototype testing, light
indicators (LEDs) are integrated to make sure of the
prototype user friendly and implementable.
The Fig. 3 describes the overall process flow of
designed frame work. It focuses on the real time
implementation to support our proposed buses
notification system. It contains several intelligent
decision support steps. The process flow starts by
taking into consideration all the aspects from initial
login system till the ending aim of user friendly display
at all slave modules.
EXPERIMENTAL ANALYSIS
The major aim of the research is to optimize one
single route’s coordinates into its single digit
Overall system integration and flow: The
representative through RSM. The techniques applied
methodology section above described the individual
have a clear potential to provide a reliable public
master and slave nodes integration, their proposed
transport monitoring system by a reduced amount of
deployment architecture and functioning. Finally here is
data to compare for MM and obviously the time (data
presenting the overall structural design and integration
rate) saving. This paper also propose and identifies a
of proposed system. The block diagram described in
unique design phase containing a system to mount GPS
Fig. 2 pictorially explains the real time hardware
based embedded system with sophisticated RF
implementable form of the system.
transceiver on every bus and at the bus station. For the
5175
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Table 4: Analysis of variance
Source
SS
Model
246.98
A-Longitude M
5.284E-003
B-Longitude S
1.11
C-Latitude M
1.113E-005
D-Latitude S
4.146E-005
Residual
0.52
Cor total
247.50
Mean Square
24.70
5.284E-003
1.11
1.113E-005
4.146E-005
0.027
F-value
906.18
0.19
40.71
4.085E-004
1.521E-003
Residual
purpose, mature prototypes are constructed presenting
master and slave modules. This section analyzes and
evaluates the main results obtained from prototype,
which will prove the potential implementation of the
proposed system. The evaluation is carried out by
observing three aims:
•
•
•
First is predicted output of the equation vs actual
observation for bus stop number by RSM
(ANOVA).
Second is equation validation through MCU for
accuracy measure.
Third is the time reduction claim during MM for 1
digit output comparison vs casual necessary 17
digits for coordinates of particular location.
Analysis of Variance (ANOVA): The presented
technique is aimed to develop the input-output
relationships for prediction of bus stop ID. In order to
arrive at the most influential variables and its effects a
phase strategies were proposed. Response surface
methodology (RSM) based on Central Composite
Design (CCD) was utilized to develop a linear model
for prediction as well as to estimate the transfer
functions at the optimal region.
The Analysis of Variance (ANOVA) results are
presented in Table 4. It can be seen that the model Fvalue of 906.18 implies the model is significant. There
is only a 0.01% chance that a "Model F-Value" could
occur. The reason behind this could be the irregular
behavior of the coordinates as input. Values of
probability less than 0.0500 indicates model terms will
produce a minor difference among the trained and
actual response.
Finally using the collected then pre processed data,
a mathematical prediction model has been developed
based on the most influencing factors and the validation
simulation analysis proved its adequacy. The result
aimed towards prediction of bus stop number and its
effect on received coordinate’s data. The von Mises and
displacement equation representing the selected route’s
coordinates is expressed in Eq. (1).
P = 25.20266832 + 1.321582163*A +
0.02467396989*B - 7.753245435*C0.1525904315*D
(1)
p-value Prob>F
< 0.0001
0.6647
< 0.0001
0.9841
0.9693
Table 5: R2 analysis
Parameter
Std. Dev.
Mean
R-Squared
Adj R-Squared
Pred R-Squared
Significant
Value
3.211308
6.000
0.980625
0.977857
0.97994
where,
P = Predicted bus stop number
A = Longitude Minutes
B = Longitute Second
C = Latitude Minute
D = Latitude Second
The Eq. (1) obtained gave a simple one digit
representing particular bus stop number. The simplicity
of the equation encouraged the researcher to use the
RSM. This present form of the equation is
uncomplicated to train the MCU using the parameters
of data. The all four parameter are programmed in the
MCU to receive as the long integer continuously.
Instead of comparing digit by digit of all location
coordinates digits, the above equation provides a single
optimized value output representing whole location
coordinates in order to trigger the RF modules for
location transmission through looping of sensor
network.
The R2 analysis results are tabulated in Table 5.
The "Pred R-Squared " 0.97994 is in reasonable
agreement with the "Adj R-Squared " of 0.977857. In
fact, when the value of correlation coefficient R is close
to 1, it means the response correlation of actual
received coordinates and predicted values are better.
The average accuracy from the R2 analysis is found out
as 97.9%. The statistical analysis shows that, the
developed mathematical model to predict the exact bus
stop, based on central composite design is statistically
adequate and can be used to trigger the slave module to
start looping of transceiver for a single bus route.
Figure 4 shows the predicted versus actual plot
how the model predicts over the range of data. The best
fit line plot of the 33 points, Table 6 is found to be close
to the ideal line (Y = X ). The predicted responses show
the good agreement with actual results. The scatter
shows the bus stop number representing casual position
data from longitude and latitude can be predicted very
precisely through RSM.
5176
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Fig. 4: Predicted versus actual plot
Table 6: Actual versus predicted (Fitted)
Observations
Actual
Fitted
1
1.00000
1.00420
2
1.00000
1.02440
3
1.00000
1.04212
4
2.00000
2.69828
5
2.00000
2.75146
6
2.00000
2.77165
7
3.00000
2.70039
8
3.00000
2.72058
9
3.00000
2.72798
10
4.00000
3.51060
11
4.00000
3.52047
12
4.00000
3.52787
13
5.00000
4.62960
14
5.00000
4.68031
15
5.00000
4.73349
16
6.00000
6.71759
17
6.00000
6.76830
18
6.00000
6.80622
19
7.00000
7.23611
20
7.00000
7.21299
21
7.00000
7.19234
22
8.00000
7.71142
23
8.00000
7.76706
24
8.00000
7.81778
25
9.00000
8.55423
26
9.00000
8.62267
27
9.00000
8.65813
28
10.0000
9.62568
29
10.0000
9.58730
30
10.0000
9.53120
31
11.0000
11.3192
32
11.0000
11.3851
33
11.0000
11.4433
Residual
-0.00420
-0.02440
-0.04212
-0.69828
-0.75146
-0.77165
0.29961
0.27942
0.27202
0.48940
0.47953
0.47213
0.37040
0.31969
0.26651
-0.71759
-0.76830
-0.80622
-0.23611
-0.21299
-0.19234
0.28858
0.23294
0.18222
0.44577
0.37733
0.34187
0.37432
0.41270
0.46880
-0.31917
-0.38514
-0.44326
Equation validation and field test: RSM results
provided a good agreement between the received and
trained coordinates in terms of single digit representing
whole location data. The novel equation obtained, gave
a simple functional dimension to train to the MCU. The
subsequent step is to embed the equation into the MCU.
In order to embed and validate the equation, MCU is
programmed such that all the constants and variables of
equation are taken as floating point. While the
necessary location coordinates are received such that
the selected digits from them can be utilized for the
variables A, B, C and D mentioned in the Eq. (1).
For the testing and equation validation purpose, the
master module prototype was brought to the exact route
of KL corresponding to the collected coordinates for
training (Table 1). The equation and modules gave well
expected results in the form of one digit representing
the exact bus stop number at each slave node
destination. During master module testing, the equation
is programmed in MCU to calculate the output up to
only one (01) digit after decimal number for
conditioning and comparison. Later on the five digits
(05) after decimals are taken as to verify the accuracy
of the system since the calculation comprise of floating
point based mathematics. This selection is inspired by
the same quantity of digits obtained by DOE using
RSM (Table 6).
Accuracy measure: The evidence is collected by
programming the MCU to obtain the results of equation
calculation step by step on the serial communicator
provided by PIC kit 2 and display at the LCD as well.
The purpose is to reach at the accuracy for the equation
validation. Although there is a very slight difference
found in the digits, (the reason is floating point
calculation capabilities by selected 8 bit MCU)
however these distinctions did not cast any effect on
calculating the one digit value representing the bus stop
number. This neglect-able error factor is also avoided
by the help of conditioning of the output of the
equation. For example for bus stop “01”, the output of
the equation by MCU “1.02” is conditioned to “1” The
condition is applied by observing the maximum limit of
the 1 digit after decimal. Therefore, it can be interpreted
Table 7: Comparison with accuracy measure for all 11 bus stops calculation results by RSM and MCU
Observations
Actual
Fitted (RSM)
Fitted (MCU)
1
1
1.0244
1.02439
2
2
2.75146
2.75144
3
3
2.72058
2.72058
4
4
3.52047
3.52046
5
5
4.68031
4.68031
6
6
6.7683
6.76829
7
7
7.21299
7.21298
8
8
7.76706
7.76705
9
9
8.62267
8.62266
10
10
9.5873
9.58729
11
11
11.3851
11.38513
5177
Error (%)
9.76E-04
7.27E-04
0.00E+00
2.84E-04
0.00E+00
1.48E-04
1.39E-04
1.29E-04
1.16E-04
1.04E-04
-2.64E-04
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
that the accuracy measure check after testing the
hardware for more than 8 digits based calculation have
no error effect on overall functionality of the system.
The Table 7 shows the overall calculation end results
comparison through DOE and MCU for all 11 bus
stops.
The Table 7 shows a comparative analysis of end
bus stop results after all calculation for all 11
observations. The remarkable results can be seen above.
In the worst condition, it is observed a deviation of
9.76E-04 by calculation for bus stop number 01.
Therefore, the overall results from RSM as well as
MCU implies that the goal to achieve the target of
obtaining single digit bus stop number representative of
17 casual digits of location data was found to be
capable efficient enough as compared to the RSM based
simulation, since the required significant bits for
conditioning is just one digit after decimal.
Data and time reduction claim: The successful data
size reduction for efficient comparison, the outcomes of
the research cast a considerable influence on limiting
the time consumption as well for MM and transmission.
This contribution is accomplished using one digit
output by the novel route equation. The one digit will
represent 17 digits necessary location data consist of
longitude and latitude. Theoretically one digit replacing
17 digits position data must have a potential to reduce
the data and time by 17 times. But practically it does
not happen so. For the purpose being, several testing
based experiments were conducted to evaluate the
actual processing time taken by the modules. Master
module is selected since it is the hub for all the
operations done in overall system.
Figure 5 shows that the processing time for master
module includes receiving, calculating and transmitting.
The objective of this finding is to estimate the time at
which PIC16F877A with 4MHz crystal performs
certain calculations since the equation contains several
floating point based calculations and to verify the
overall time reduction claim. We managed to compute
the processing time period during three steps through
digital oscilloscope GDS 1062A along with its serial
simulator to capture on laptop as well as PICkit2 logic
analyzer tool (LA).
Initially it was checked either the system is free of
error or not. For the purpose, a simple delay of 100ms
and 50ms routine was made and verified through the
oscilloscope and logic analyzer tool using the digital
output and cursor feature respectively at the output pin
of PIC using the blinking LED. The code displayed a
square wave with the period cycle of 200ms with
100ms delay and 100ms with 50ms delay. It was
observed no error in our procedure. Later on, actual
master module processing tasks for time measurement
was checked through the change of state by LED.
As mentioned earlier, the processing spectra
contain three tasks that is receiving, computing and
transmitting. The code was written to do all three tasks
one by one in loop. Three LED are selected for each
tasks. Before every task, the respective LED was made
HIGH and, LOW after finishing particular task. Since
Fig. 5: Overall processing tasks in time sequence
Fig. 6: Time taken by MCU to receive GPS selected data (Longitude, Latitude, valid and direction character) by oscilloscope
5178
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Fig. 7: Time taken by MCU for all floating point calculation using LA
Fig. 8: Time taken by MCU for transmitting one digit
Fig. 9: Time taken by MCU for transmitting 17 necessary digits
the duration between them to toggle will help to
estimate the actual time consumption. The output was
observed at the oscilloscope as well as logic analyzer.
For the first task, location data reception was found
unable to measure accurately by LA. For the purpose,
the oscilloscope was utilized.
The Fig. 6 shows the time consumption for
receiving specific location data by MCU as master
module. As mentioned above, in order to capture the
duration, LED1 output is programmed to toggle after
receiving GPS data, is captured at the channel 1 for
both LA and oscilloscope. LA was found having
insufficient range to compute the time taken by MCU.
Furthermore, the same procedure was followed using
oscilloscope as an individual step. Reception is initiated
with a high pulse and ended by low one, but the total
time period is taken for one complete cycle of period,
which is found to be 1500ms as shown in the figure.
These results are taken as average by capturing the
reading after 5 run by the oscilloscope. It implies the
average actual time taken by MCU to receive the
selected GPS data took ~1500/2=~750ms.
Figure 7 shows the time consumption for the
floating point based calculation inside master module.
In order to capture the duration, an LED2 output is
programmed to toggle after completion of all
calculation until gives the one digit bus stop number, is
captured at the channel 2 using LA. In the programming
sequence, this task is kept right second after GPS
reception. This can be evidence by the figure, that the X
for calculation time starts right after GPS reception
pulse fall down at channel 1. The duration is estimated
about ~1.3ms.
Figure 8 shows the time consumption for
transmitting one digit output from the calculation. The
one digit output is the bus stop number sent serially to
the prototype slave node. In order to capture the
duration, an LED3 output is programmed to toggle after
transmitting the digit representing minimum 17digits of
location data, at channel 3 using LA. However as
mentioned earlier, the selected LA was found incapable
of capturing 3 channels at a time. This problem is
resolved by the programming sequence. In the
sequence, this task is kept right third after calculation
complete (2nd task). The time taken was easily captured
from falling edge of 2nd task until the rising edge of 1st
task, as evidence by the figure. The duration is
estimated about ~4.92ms.
The Fig. 9 shows the time consumption for
transmitting 17 necessary digits output prior to
calculation. It was captured through LA separately in
order to verify the claim of saving the time and data
5179
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Table 8: Processing time measurement summery
Operation
Captured estimated time
GPS Location reception
~750ms
All floating points calculations
1.3ms
One digit transmission
4.92ms
17 digits transmission
28.32ms
consumption. The figure clearly shows that the
minimum time taken by 17 digits to transmit serially
through the MCU is found to be 28.32ms. This time is
about 6 times higher than the time taken by 1 digit. The
Table 8 summarizes the overall time taken for the
selected tasks.
The conclusion of the task is to verify time taken
for transmitting/ receiving 17 digits vs. 1 digit. The
time reduction is found as 5.75 time more than formal
GPS data MM and transmitting. The table clearly
shows the time reduction objective was achieved during
transmitting the one digit bus stop number from master
node and receiving, comparing and transmitting
through each slave modules. The major portion of the
time taken by overall operation was coordinates
reception. Since this operation is not only the function
of just 17 digits reception but also several other
necessities such as waiting for valid variable “A” and
distinction amongst the degree, minutes and seconds by
waiting for their proper sign of “.” and “,” etc.
DISCUSSION ON OPERATIONAL
EFFECTIVENESS
The overall performance and implementation for
proposed study at individual bus companies is
determined by its technical as well as organizational
factors too. Technically the implementation of the
provided solution in the field of the intelligent
transportation system seems a huge task since it
required a profound cooperation of the transport service
provider and the urban management system. So during
the prototype designing, there are several tribulations
faced, one of them is the delay in receiving the position
data by GPS and second one can be the node failure for
the wireless sensor systems at Buses' station. Therefore,
these conditions should be taken into account that the
whole system can work properly, even if some of the
nodes do not work temporarily. According to looping
routing mechanism designed using RF modules, the
basic idea ensuring the reliability of communication is:
the unique ID of bus must transmit properly, the main
path must be created first from the master module to the
slave module one by one. By keeping above scenario,
this research adopts an individual as well distributed
looping algorithm. That is, the communication is
utilized in the way to have a greater functioning
capability in the network for the Intelligent
Transportation System.
CONCLUSION
the
This study presents an approach that includes both
individual bus route optimization and the
technological utilization to provide the public transport
an implementable structure in the wider aspect. MM
algorithm through RSM for better and reliable route’s
coordinate
optimization
brought
remarkable
performance evaluation to be used as public transport
notification system in terms of data and time reduction.
In case decision support system, the generic needs of
the user, such as updated information with reliability
are well documented in the literature and are
subsequently reinforced in this study up to more
technological enhancement. The design of the
framework impacts the fruitful investment in long term
to the companies operating the services. The model
presented, illustrates several key stages which have to
be taken into account by the transport planner. Anyhow,
successful designing, its optimization, designed results
matching, equation validation and testing directs a clear
potential to bring the system in the leading solutions to
resolve the overall public transport monitoring system
for passengers. After testing prototype and
modifications in the documentation for about a year by
taking extensive field test and surveys, the proposed
framework tends to have potential for commanding
public transit operations and lend a hand to change the
trend of passengers to switch to use public transport
instead of private own cars. And finally the reliable
solution for the traffic congestion by providing
facilitated bus stops to the travelers.
As far as the future works concern, our focus will
be on designing of centrally monitoring server based
system by which the positioning data base of all the
tracked vehicles could be examined. This would be help
full to realize the driver efficiency as well as the factors
effecting on well-organized monitoring system. For the
purpose being, the GSM technology will be suggested
to utilize.
REFERENCES
Abd Alla, A.N., M. Rauf, A. Fakharuddin and N.M.K.
Nik yousuff, 2011. A novel identification and
monitoring technique of multi ID public transport
for station reporting. Sci. Res. Essays, 6(10):
2129-2136.
Ban, X., P. Hao and Z. Sun, 2011. Real time queue
length estimation for signalized intersections using
travel times from mobile sensors. Trans. Res. Part
C Emerg. Technol., 19(6): 1133-1156.
Bao, Y., W. Guojiang and Y. Jiaming, 2002. Overall
design of Xiamen City GPS intelligent vehicle
monitoring system. Proceedings of the IEEE,
International Conference on Control and
Automation, pp: 887-891.
Bong, G.K., H.S. Yong, H.A. Seon, T.C. Kyung and
G.L. Bong, 2007. Building BIS/BMS with wireless
communication system in Korea. Proceeding of
IEEE 6th International Conference on Advanced
Language Processing and Web Information
Technology, pp: 376-380.
5180
Res. J. Appl. Sci. Eng. Technol., 5(22): 5170-5181, 2013
Ezell, S., 2010. Explaining international IT application
leaderhip: Intelligent transportation systems. The
Information
Technology
and
Innovation
Foundation (ITIF), Tech. Rep.
Fan, X. and F. Jiancheng, 2007. Velocity and position
error compensation using strapdown inertial
navigation system/celestial navigation system
integration based on ensemble neural network.
Aerospace Sci. Technol., 111(4): 222-228.
GR Reporter, 2011. 52 Million Euros for Installing
"smart" Bus Stops in Athens, the Debts of Public
Transport are Growing, Retrieved from:
http://www.grreporter.info/en,
(Accessed
on:
March 2nd, 2011).
Herrera, J.C., D.B. Work, R. Herring, X. Ban, Q.
Jacobson and A.M. Bayen, 2010. Evaluation of
traffic data obtained via GPS-enabled mobile
phones: The mobile century field experiment.
Transport. Res. Part C, 18: 568-583.
Lee, S.Y. and S.W. Han, 2003. Geospatial construction
data collection using GPS. KSCE J. Civil Eng.,
7(4): 363-370.
Liu, L. and M.G. Amin, 2009. Tracking performance
and average error analysis of GPS discriminators in
multipath. Signal Proc., 89(6): 1224-1239.
Lu, M., W. Chen, X. Shen, H.C. Lam and J. Liu, 2007.
Positioning and tracking construction vehicles in
highly dense urban areas and building construction
sites. Automat. Construc., 16: 647-656.
Mintsis, G., S. Basbas, P. Papaioannou, C. Taxiltaris
and I.N. Tziavos, 2004. Applications of GPS
technology in the land transportation system.
Europ. J. Oper. Res., 152: 399-409.
Niu, H., W. Guan and J. Ma, 2009. Design and
Implementation of bus monitoring system based on
GPS for Beijing Olympics. Proceeding of WRI
World Congress on Computer Science and
Information Engineering. Los Angeles, CA, 7:
540-544.
Noureldin, A., A. El-Shafie and M. Bayoumi, 2011.
GPS/INS integration utilizing dynamic neural
networks for vehicular navigation. Inform. Fusion,
12(1): 48-57.
Qin, K., J. Xing, G. Chen, L. Wang and J. Qin, 2008.
The design of intelligent bus movement monitoring
and station reporting system. Proceedings of the
IEEE, International Conference on Automation and
Logistics. China, pp: 2822-2827.
Rauf, M., A.N. Abd Alla, K. Hawari bin Ghazali and
A. Fakharuddin, 2010. Intelligent multi ID buses
navigation system with efficient data transmission
technique. Proceedings of the IEEE International
Conference on Information Networking and
Automation (ICINA), 2010, 1: V1-57-V161.
Raymond, H.M. and C.M. Douglas, 2002. Response
Surface Methodology: Process and Product
Optimization using Designed Experiments. 2nd
Edn., John Wiley and Sons, USA.
Rowland, H. and J. Antony, 2003. Application of
design of experiments to a spot welding process.
Assemb. Autom., 23(3): 273-279.
Sharaf, R. and A. Noureldin, 2007. Sensor integration
for satellite-based vehicular navigation using
neural networks. IEEE T. Neural Networ., 18(2):
589-594.
Soo, T.S.T., C.T. Han and T.A. Rahman, 2007.
Intelligent fleet management system with
concurrent GPS & GSM real-time positioning
technology. Proceeding of 7th International
Conference on ITS Telecommunications (ITST),
pp: 1-6.
United States Government Accountability Office
(GAO), 2009. Surface transportation: Efforts to
address highway congestion through Real-time
traffic information systems are expanding but face
implementation
challenges.
Report
to
Congressional Requesters. GAO-10-121R.
Xuesong, S., C. Wu and L. Ming, 2008. Wireless sensor
networks for resources tracking at building
construction sites. Tsing. Sci. Technol., 13(S1):
78-83.
5181